OBJECTIVE: We conducted a retrospective data analysis to review the results of robot-assisted arm rehabilitation in post stroke patients during past 2 years and find out positive influences of the outcomes. METHOD: We measured improvements of arm function longitudinally in a group of sixty-four stroke patients, who participated in the robot-assisted arm therapy from January 2012 to December 2013. Treatment session lasted 30 to 40 minutes, 2 to 5 times a week. For at least more than one month, we used the InMotion2.0 (Interactive Motion Technologies, Watertown, MA, USA) and measured outcomes with the Fugl-Meyer assessment-upper extremity (FMA-UE), Korea-modified Barthel index (K-MBI) and InMotion robot arm evaluation index. Also, analysis on the subgroup was carried out. RESULTS: Following the robot-assisted arm rehabilitation, FMA-UE, K-MBI and InMotion robot arm evaluation index were significantly improved compared to baseline. Mean FMA-UE and K-MBI gain were 4.22 +/- 0.76, 7.63 +/- 1.18 in each. However, in the subgroup analysis, the group with less intensity treatment (640 repetition) did not show any significant improvement. CONCLUSION: This is an observational study showing improvements in arm function following robot-assisted arm rehabilitation compared to baseline, which was significant only in the subgroup who received the intervention longer and more intensity.